A low-power integrated data processing system and method based on cable tunnel

Abstract

The present invention discloses a low-power integrated data processing system and method based on a cable tunnel, belonging to the field of artificial intelligence technology. The present invention monitors the battery status of each device in the cable tunnel in real time, judges the remaining battery power in combination with the battery life, and triggers a low-power mode when the power drops to a set threshold; controls the power supply of the cable tunnel through PMIC; processes events in an interrupt-driven manner, and enters a low-power mode after the processing is completed; uses a lightweight algorithm for preliminary data processing and screening, discards unnecessary data, and uses a low-power wireless communication protocol for data transmission; prefabricates the key parameters of the system through software tools, and performs pre-configuration, and the pre-configured content includes sensor calibration values ​​and data processing thresholds; uses a decision tree algorithm to analyze the status information of each device in the tunnel, outputs a power supply strategy for the low-power mode, and applies it to the low-power mode of the PMIC.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a low-power integrated data processing system and method based on a cable tunnel. Background Art

[0002] In modern cable tunnel management systems, data acquisition and monitoring are key links to ensure safe tunnel operation. These systems usually include a variety of sensors, monitoring devices, and communication modules to monitor the operating status of cables, environmental conditions, and the health of tunnel structures in real time. As the scale of cable tunnels increases and the demand for monitoring increases, the number and complexity of devices in the system have also increased significantly. These devices usually require continuous power supply to ensure the continuity of data acquisition and the stability of the system. However, battery-powered devices are easily limited by battery life and power during long-term operation.

[0003] The existing cable tunnel monitoring system usually adopts the traditional power supply management method during data collection and transmission, and cannot dynamically adjust the power supply strategy according to the actual working status and environmental conditions of the equipment. This leads to a lot of energy waste, especially when the equipment is in idle state, it still consumes a high amount of electricity. The current cable tunnel monitoring system lacks an intelligent power management mechanism, and the system cannot adjust the power supply mode according to real-time data. This situation will cause the system to remain in high energy consumption mode during low load or non-task periods, reducing the equipment's efficiency and battery life. Summary of the invention

[0004] The object of the present invention is to provide a low-power integrated data processing system and method based on a cable tunnel to solve the problems raised in the above background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] A low-power integrated data processing method based on a cable tunnel, the method comprising the following steps:

[0007] S100, using a battery power monitoring device to monitor the battery status of each device in the cable tunnel in real time, and judging the remaining battery power in combination with the battery life, and triggering a low power mode when the power drops to a set threshold; controlling the power supply of the cable tunnel through the power management integrated circuit PMIC, using a reinforcement learning algorithm to adjust the power supply of each device in the cable tunnel according to demand, and entering a low power mode when there is no task;

[0008] S200, process events in an interrupt-driven manner. When data is collected or an important event occurs, the system is awakened through the interrupt mechanism, and enters a low-power mode after the processing is completed; a lightweight algorithm is used for preliminary data processing and screening, and unnecessary data is discarded to reduce power consumption for data transmission and storage; a low-power wireless communication protocol LoRa is used for data transmission to further reduce power consumption;

[0009] S300, in the manufacturing stage, prefabricate key parameters of the system through software tools, wherein the key parameters are preconfigured according to the specific environment and application requirements of the cable tunnel, and the preconfigured contents include sensor calibration values ​​and data processing thresholds;

[0010] S400, using a decision tree algorithm, by analyzing the status information of each device in the tunnel, determining whether to shut down or reduce the power supply of certain devices to reduce overall energy consumption, outputting a power supply strategy for a low power mode, and applying it to the low power mode of the PMIC.

[0011] According to step S100, the battery power monitoring device is responsible for continuously monitoring the battery status of each device in the cable tunnel, collecting the output voltage of the battery through the analog-to-digital converter, and the voltage gradually decreases as the battery discharge process proceeds; the current flowing through the battery is collected through the current sensor, and the actual discharge amount of the battery in a given time period is calculated by the coulomb counting method;

[0012] The monitoring device records the initial capacity of the battery. As the battery is used, the monitoring device adjusts the actual available capacity of the battery by monitoring the capacity decay during the charge and discharge process. The monitoring device calculates the remaining capacity using the coulomb counting method based on the current voltage, current and the actual available capacity of the battery.

[0013] During the initial use of the battery, the monitoring device records its maximum capacity, that is, the maximum energy that the battery can store in its new state. As the battery is used and over time, the actual capacity of the battery will gradually decay. The monitoring device determines the degree of battery capacity decay by monitoring the actual energy that the battery can accommodate and release during each charge and discharge process, and comparing it with the initial capacity. Through repeated monitoring and recording, the monitoring device can update the actual available capacity of the battery in real time and use it as a new benchmark for evaluating the battery's health status and remaining power.

[0014] During real-time monitoring, the monitoring device continuously tracks the changes in the battery's voltage and current. When the battery is in a discharging state, the monitoring device records the amount of charge released by the battery; when the battery is in a charging state, it records the amount of charge absorbed by the battery. Combined with the actual available capacity adjusted previously, the monitoring device can calculate the current remaining capacity of the battery by accumulating these charges. The principle of the coulomb counting method is to accurately calculate the amount of electricity flowing out of or into the battery during the entire battery life, thereby accurately estimating the remaining capacity of the battery.

[0015] When the battery power drops to a preset warning threshold, the battery power monitoring device generates a status feedback signal and sends the status feedback signal to the power management integrated circuit PMIC to trigger a low power consumption mode.

[0016] According to step S100, the PMIC is responsible for unified management of the power supply of each device. The PMIC uses a reinforcement learning algorithm to establish a reinforcement learning model based on the operating status and actual needs of different devices in the tunnel, and dynamically adjusts the power supply strategy. The PMIC monitors the operating status of the devices in the tunnel and identifies the current working mode. The PMIC evaluates the power demand of each device based on the power supply strategy and determines the required power supply level. Based on the evaluation results of the device demand, the PMIC adjusts the output voltage and current. The PMIC provides corresponding power support when the device has an actual task. During the non-task period, the PMIC turns off the power supply of the module of the corresponding device and enters a low power consumption mode.

[0017] PMIC assigns power supply priorities to different devices. There are several power inputs in the cable tunnel, and PMIC is responsible for switching between power supplies. PMIC has redundant power management capabilities. When a power supply fails, PMIC can switch to another power path to ensure uninterrupted power supply to the device.

[0018] When allocating power, the PMIC will determine the power supply priority of each device based on the importance, mission requirements, and current working status of each device in the cable tunnel. Critical devices, such as sensors or communication modules for security monitoring, will have a higher priority to ensure that they can operate stably under any circumstances. Next are devices that need to collect data regularly. The power supply priority of these devices can be dynamically adjusted according to the urgency of the task. For those devices that are not frequently used or are not important in the current task, the PMIC will lower their power supply priority or even completely cut off the power supply when it is not necessary, thereby reducing the energy consumption of the entire system.

[0019] In cable tunnels, due to the complexity of the environment and the importance of the equipment, multiple power inputs are usually set up to ensure uninterrupted power supply to the equipment. PMIC has an intelligent power switching function. When it detects that a certain power supply fails or cannot supply power normally, PMIC can automatically switch to other backup power paths. This switching is seamless, ensuring that the operation of the equipment will not be interrupted due to power failure. PMIC determines the current power supply status by monitoring the status of each power input, such as voltage, power supply stability, etc. Once an abnormality is detected, PMIC will give priority to the power supply in good health, and will also report the fault information to the system for further maintenance and inspection. Through this redundant power management, PMIC ensures continuous and stable power supply to all equipment in the cable tunnel, avoiding system downtime caused by single power failure.

[0020] The reinforcement learning model includes a state space, an action space and a reward function. The state space represents the system state that the PMIC can observe, including the operating state, battery state and environmental state of each device, wherein the operating state includes working, standby and sleep, the battery state includes the power percentage, voltage and current, and the environmental state includes temperature and humidity; the action space represents the actions that the PMIC can take, including high-power power supply, low-power power supply and power off, and the high-power power supply and low-power power supply are specifically adjusted according to the state space; the reward function is used to measure the effect of the action taken by the PMIC, and the reward value is associated with the energy consumption and performance goals of the system;

[0021] The Q-Learning algorithm is used as a reinforcement learning method to determine the strategy by iteratively updating a table called the Q value. The formula is as follows:

[0022] ;

[0023] in, is in state Take Action The Q value at the time represents the execution action The expected cumulative reward after is the learning rate, ranging from , control the degree to which new information updates old information; Is the immediate reward at the current moment, reflecting the action effect; is the discount factor, ranging from , weighing the relative importance of future rewards versus current rewards; Indicates that in the next state In the example above, choose the action that may bring the greatest reward. The corresponding Q value;

[0024] PMIC continuously monitors the real-time status of the equipment in the tunnel and uses the real-time status as the input S of the reinforcement learning algorithm. At each time step t, PMIC And the learned Q value, choose an optimal action ;

[0025] When the PMIC performs an action, the system gives an instant reward based on the current energy consumption and device performance. ,When the PMIC successfully reduces unnecessary power supply and the device still operates normally, the system gives a positive reward; if the insufficient power supply causes the device to degrade, a negative reward is given;

[0026] PMIC continuously updates the Q value table through the Q-learning formula and gradually learns the optimal power management strategy under various conditions.

[0027] According to step S200, in the non-task or standby state, the system's PMIC puts the system into a low-power mode, turns off or reduces the power supply of unnecessary modules to minimize energy consumption; when the preset conditions are met, an interrupt signal is triggered, and the preset conditions include the sensor detecting a specific signal, an external input request, and the expiration of a timer; the interrupt signal interrupts the current low-power mode of the system and wakes up the processor; after the system responds to the interrupt signal, the processor enters the interrupt service program, which is responsible for handling the specific event that triggers the interrupt, and the system can complete data collection and data processing in the ISR; when the interrupt service program is executed, the system will return to the previous low-power mode; the PMIC will turn off the power supply of modules that are no longer needed, and the processor will enter a sleep or standby state, waiting for the next interrupt trigger.

[0028] According to step S200, the data collected by the sensor contains redundant information and noise. The lightweight algorithm preprocesses the raw data, including denoising, normalization and feature extraction, and uses a moving average algorithm for denoising. In the data screening stage, the threshold filtering algorithm and the differential algorithm are used to classify the data according to preset conditions or thresholds. Data that meet specific conditions will be retained, and the remaining data will be marked as non-essential data and discarded. The specific conditions represent abnormal data that exceeds the set range or key data within a specific time period. For the data of the temperature and humidity sensors, fluctuations within the normal range are regarded as non-essential data and are discarded. When the temperature or humidity exceeds the set range, the data will be retained and transmitted. For the monitoring data of the operating status of certain equipment in the cable tunnel, when the equipment is working in a normal state, the generated data is discarded; when the equipment is abnormal, the generated data will be retained.

[0029] The threshold filtering algorithm retains data that exceeds the threshold range and discards data within the normal range by setting one or more thresholds; the differential algorithm screens out data points with significant changes by calculating the difference between the current data and the data at the previous moment, and retains the data with significant changes. The threshold for determining whether the change is significant is determined by pre-configured content.

[0030] According to step S300, the sensor calibration value refers to the parameter obtained by calibrating the sensor according to the standard reference environment during the production process, which is used to ensure the measurement accuracy of the sensor in actual use; in the production workshop, the environmental conditions of the cable tunnel are simulated to calibrate the sensor; according to the calibration result, the calibration parameters of the sensor are adjusted using a software tool, and for different sensor types, the relationship between the sensor output and the actual physical quantity is modeled by a linear regression algorithm to determine the calibration value; the temperature sensor is calibrated by a standard source with a known temperature to obtain the calibration value;

[0031] The data processing threshold is a parameter used by the system to judge and filter data during operation, which determines when the system takes further processing or alarm operations on the collected data. In the production stage, the threshold range of different types of data is determined based on the historical data of the cable tunnel or the results of environmental simulation. The threshold is written into the system configuration file or firmware using software tools. The system can judge the validity or abnormality of the data during actual operation based on the pre-configured threshold.

[0032] According to step S400, the input features of the decision tree include the current status and environmental parameters of each device in the cable tunnel. The input features are used to construct the decision tree model, including: the battery power of the device , the current task load of the device , Equipment ambient temperature , the current working mode of the device and device task priorities ;

[0033] Each decision node makes a binary judgment based on a feature to decide which child node to move to. The input feature to Corresponding to nodes 1 to 5 respectively, each node determines whether the corresponding input feature is higher or lower than the set threshold, and the judgment is performed from node 1 to node 5 in sequence;

[0034] The leaf node represents the final power supply strategy decision. The action options of the leaf node include shutting down the device, reducing the power supply, and maintaining the power supply. If the battery power of the device is lower than the set threshold, and the current task load of the device is lower than the set threshold, and the ambient temperature of the device is lower than the set threshold, the device is shut down to save power; if the battery power of the device is lower than the set threshold but the task load still exists, the power supply voltage is reduced; if the current task load of the device is higher than the set threshold or the device is executing a high-priority task, the current power supply state is maintained.

[0035] According to step S400, historical data is used to mark the power supply strategy of each device in different states, so as to train the decision tree model; the best splitting feature of each decision node is selected by using information gain, and the feature space is recursively split until the preset node purity is reached; according to the current state of the device, the optimal power supply strategy is obtained through the decision tree model; in order to prevent the decision tree model from overfitting, the tree is pruned and unnecessary branches are removed; according to the feedback data in actual operation, the decision tree model is continuously updated and optimized.

[0036] A low-power integrated data processing system based on a cable tunnel, comprising:

[0037] Battery monitoring and low-power trigger module: including: battery power monitoring unit, low-power trigger unit, PMIC control unit and reinforcement learning power optimization unit; the battery power monitoring unit uses a battery power monitoring device to monitor the battery status of each device in the cable tunnel in real time, the low-power trigger unit determines the remaining battery power in combination with the battery life, and triggers the low-power mode when the power drops to the set threshold. The PMIC control unit controls the power supply of the cable tunnel through the power management integrated circuit PMIC, and the reinforcement learning power optimization unit uses the reinforcement learning algorithm to adjust the power supply of each device in the cable tunnel according to demand, and enters the low-power mode when there is no task;

[0038] Event processing and data transmission optimization module: including: interrupt drive unit, data processing and screening unit and LoRa communication unit; the interrupt drive unit processes events by interrupt drive, wakes up the system through the interrupt mechanism when data is collected or important events occur, and enters low power mode after processing. The data processing and screening unit uses a lightweight algorithm to perform preliminary data processing and screening, discards unnecessary data, and reduces the power consumption of data transmission and storage. The LoRa communication unit uses the low-power wireless communication protocol LoRa for data transmission, further reducing power consumption;

[0039] Parameter prefabrication and configuration module: including: parameter prefabrication unit and parameter configuration unit; wherein, the parameter prefabrication unit prefabricates the key parameters of the system through software tools during the manufacturing stage, and the key parameters are pre-configured by the parameter configuration unit according to the specific environment and application requirements of the cable tunnel, and the pre-configured contents include sensor calibration values ​​and data processing thresholds;

[0040] Low power strategy decision module: including: a decision tree analysis unit and a low power strategy output unit; wherein, the decision tree analysis unit uses a decision tree algorithm to analyze the status information of each device in the tunnel to determine whether to shut down or reduce the power supply of certain devices to reduce overall energy consumption, and the low power strategy output unit outputs the power supply strategy of the low power mode, which is applied to the low power mode of the PMIC.

[0041] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0042] The present invention introduces an intelligent power management strategy based on PMIC. Through the reinforcement learning algorithm, the system can dynamically adjust the power supply strategy according to the real-time working status and task requirements of the device, and automatically switch to a low-power mode when idle, thereby significantly reducing energy consumption;

[0043] The present invention monitors the battery status of the device in real time through a battery power monitoring device, and determines the remaining power in combination with the battery life; when the power drops to a set threshold, the system can automatically trigger a low power consumption mode, thereby extending the battery life and reducing the frequency of manual maintenance;

[0044] The present invention uses a decision tree algorithm to intelligently determine whether to shut down or reduce the power supply of unnecessary modules according to the status information of the equipment in the tunnel to minimize energy consumption. This intelligent decision-making based on real-time data enables the system to achieve optimal energy consumption performance under various working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0046] Figure 1 It is a schematic diagram of the steps of a low-power integrated data processing method based on a cable tunnel of the present invention;

[0047] Figure 2 It is a system structure diagram of a low-power integrated data processing system based on a cable tunnel of the present invention. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] See also Figure 1 and Figure 2 , the present invention provides a technical solution:

[0050] According to one embodiment of the present invention, Figure 1 A schematic diagram of a low-power integrated data processing method based on a cable tunnel is shown in the following steps:

[0051] S100, using a battery power monitoring device to monitor the battery status of each device in the cable tunnel in real time, and judging the remaining battery power in combination with the battery life, and triggering a low power mode when the power drops to a set threshold; controlling the power supply of the cable tunnel through the power management integrated circuit PMIC, using a reinforcement learning algorithm to adjust the power supply of each device in the cable tunnel according to demand, and entering a low power mode when there is no task;

[0052] S200, process events in an interrupt-driven manner. When data is collected or an important event occurs, the system is awakened through the interrupt mechanism, and enters a low-power mode after the processing is completed; a lightweight algorithm is used for preliminary data processing and screening, and unnecessary data is discarded to reduce power consumption for data transmission and storage; a low-power wireless communication protocol LoRa is used for data transmission to further reduce power consumption;

[0053] S300, in the manufacturing stage, prefabricate key parameters of the system through software tools, wherein the key parameters are preconfigured according to the specific environment and application requirements of the cable tunnel, and the preconfigured contents include sensor calibration values ​​and data processing thresholds;

[0054] S400, using a decision tree algorithm, by analyzing the status information of each device in the tunnel, determining whether to shut down or reduce the power supply of certain devices to reduce overall energy consumption, outputting a power supply strategy for a low power mode, and applying it to the low power mode of the PMIC.

[0055] According to step S100, the battery power monitoring device is responsible for continuously monitoring the battery status of each device in the cable tunnel, collecting the output voltage of the battery through the analog-to-digital converter, and the voltage gradually decreases as the battery discharge process proceeds; the current flowing through the battery is collected through the current sensor, and the actual discharge amount of the battery in a given time period is calculated by the coulomb counting method;

[0056] The monitoring device records the initial capacity of the battery. As the battery is used, the monitoring device adjusts the actual available capacity of the battery by monitoring the capacity decay during the charge and discharge process. The monitoring device calculates the remaining capacity using the coulomb counting method based on the current voltage, current and the actual available capacity of the battery.

[0057] When the battery power drops to a preset warning threshold, the battery power monitoring device generates a status feedback signal and sends the status feedback signal to the power management integrated circuit PMIC to trigger a low power consumption mode.

[0058] According to step S100, the PMIC is responsible for unified management of the power supply of each device. The PMIC uses a reinforcement learning algorithm to establish a reinforcement learning model based on the operating status and actual needs of different devices in the tunnel, and dynamically adjusts the power supply strategy. The PMIC monitors the operating status of the devices in the tunnel and identifies the current working mode. The PMIC evaluates the power demand of each device based on the power supply strategy and determines the required power supply level. Based on the evaluation results of the device demand, the PMIC adjusts the output voltage and current. The PMIC provides corresponding power support when the device has an actual task. During the non-task period, the PMIC turns off the power supply of the module of the corresponding device and enters a low power consumption mode.

[0059] PMIC assigns power supply priorities to different devices. There are several power inputs in the cable tunnel, and PMIC is responsible for switching between power supplies. PMIC has redundant power management capabilities. When a power supply fails, PMIC can switch to another power path to ensure uninterrupted power supply to the device.

[0060] When allocating power, the PMIC will determine the power supply priority of each device based on the importance, mission requirements, and current working status of each device in the cable tunnel. Critical devices, such as sensors or communication modules for security monitoring, will have a higher priority to ensure that they can operate stably under any circumstances. Next are devices that need to collect data regularly. The power supply priority of these devices can be dynamically adjusted according to the urgency of the task. For those devices that are not frequently used or are not important in the current task, the PMIC will lower their power supply priority or even completely cut off the power supply when it is not necessary, thereby reducing the energy consumption of the entire system.

[0061] In cable tunnels, due to the complexity of the environment and the importance of the equipment, multiple power inputs are usually set up to ensure uninterrupted power supply to the equipment. PMIC has an intelligent power switching function. When it detects that a certain power supply fails or cannot supply power normally, PMIC can automatically switch to other backup power paths. This switching is seamless, ensuring that the operation of the equipment will not be interrupted due to power failure. PMIC determines the current power supply status by monitoring the status of each power input, such as voltage, power supply stability, etc. Once an abnormality is detected, PMIC will give priority to the power supply in good health, and will also report the fault information to the system for further maintenance and inspection. Through this redundant power management, PMIC ensures continuous and stable power supply to all equipment in the cable tunnel, avoiding system downtime caused by single power failure.

[0062] In this embodiment, multiple key devices are arranged in a 500-meter-long cable tunnel, including security monitoring sensors, communication modules, data acquisition devices, and environmental monitors. Two independent power inputs are set, each with a power of 5kW and a supply voltage of 220V AC. A PMIC chip with intelligent power priority allocation and redundant power management functions is equipped to control the power supply of all devices.

[0063] Under normal power supply conditions, test the power consumption, operating status, and task execution efficiency of each device, and record the power supply priority assigned by the PMIC. Simulate changes in task requirements, adjust the task urgency of certain devices, and test the ability of the PMIC to dynamically adjust the power supply priority. Simulate a power input failure, test the redundant power management function of the PMIC, and record the operating stability of the device after switching power.

[0064] The power consumption and priority allocation results under normal power supply conditions are as follows: Security monitoring sensor: average power consumption: 12W, power supply priority: high, operating status: stable; Communication module: average power consumption: 8W, power supply priority: high, operating status: stable; Data acquisition device: average power consumption: 15W, power supply priority: medium, operating status: stable; Environmental monitor: average power consumption: 5W, power supply priority: low, operating status: stable;

[0065] Dynamic adjustments are made after the task requirements change. When performing emergency tasks, the data acquisition device: adjusted power consumption: 20W, power supply priority: increased to high, operating status: stable, task execution efficiency increased by 30%; environmental monitor: adjusted power consumption: 3W, power supply priority: further reduced, operating status: stable, no task demand decreased.

[0066] For redundancy management in case of power failure, a fault simulation is performed: one power input fails, and the power supply is switched. The switching time is 100ms. After the switch, all devices are running normally without obvious fluctuations.

[0067] System stability: runs smoothly without interruption.

[0068] The reinforcement learning model includes a state space, an action space and a reward function. The state space represents the system state that the PMIC can observe, including the operating state, battery state and environmental state of each device, wherein the operating state includes working, standby and sleep, the battery state includes the power percentage, voltage and current, and the environmental state includes temperature and humidity; the action space represents the actions that the PMIC can take, including high-power power supply, low-power power supply and power off, and the high-power power supply and low-power power supply are specifically adjusted according to the state space; the reward function is used to measure the effect of the action taken by the PMIC, and the reward value is associated with the energy consumption and performance goals of the system;

[0069] The Q-Learning algorithm is used as a reinforcement learning method to determine the strategy by iteratively updating a table called the Q value. The formula is as follows:

[0070] ;

[0071] in, is in state Take Action The Q value at the time represents the execution action The expected cumulative reward after is the learning rate, ranging from , control the degree to which new information updates old information; Is the immediate reward at the current moment, reflecting the action effect; is the discount factor, ranging from , weighing the relative importance of future rewards versus current rewards; Indicates that in the next state In the example above, choose the action that may bring the greatest reward. The corresponding Q value; in this embodiment, the learning rate Set to 0.1, the discount factor Set it to 0.9, initialize the Q value table, where the state space is S, the action space is A, and each SA corresponds to an initial Q value of 0.

[0072] PMIC continuously monitors the real-time status of the equipment in the tunnel and uses the real-time status as the input S of the reinforcement learning algorithm. At each time step t, PMIC And the learned Q value, select an optimal action ; PMIC observes the current state and selects an optimal action based on the current Q value table. The action is selected based on Strategy, that is, based on probability Choose a random action to The probability of selecting the action with the highest Q value is . After executing the action, the system enters a new state and feedbacks the immediate reward.

[0073] When the PMIC performs an action, the system gives an instant reward based on the current energy consumption and device performance. ,When the PMIC successfully reduces unnecessary power supply and the device still operates normally, the system gives a positive reward; if the insufficient power supply causes the device to degrade, a negative reward is given;

[0074] PMIC continuously updates the Q value table through the Q-learning formula and gradually learns the optimal power management strategy under various conditions.

[0075] After 1000 iterations, the following is the policy generated by PMIC through reinforcement learning:

[0076] In state 1: Device A is in working state, battery power is 80%, temperature is moderate (25°C), humidity is 50%; optimal action: high power supply. Reason: Device A needs sufficient power supply in working state, environmental conditions are good, battery power is sufficient, and power saving is not required.

[0077] In state 2: Device B is in standby mode, battery power is 50%, temperature is high (35°C), humidity is 70%; optimal action: low power supply. Reason: Device B is in standby mode, does not need high power supply, and the temperature is high. Low power consumption helps to extend battery life.

[0078] In state 3: Device C is in sleep mode, battery power is 20%, temperature is low (10°C), humidity is 30%; optimal action: turn off the power. Reason: Device C is in sleep mode, battery power is low, and power is not needed at present. Turning off the power can save power resources.

[0079] In state 4: Device D is in working state, battery power is 10%, temperature is moderate (20°C), humidity is 40%; optimal action: low power supply. Reason: Although device D needs to work, the battery power is extremely low, so low power supply is preferred to extend the operating time and ensure that critical tasks are completed as much as possible.

[0080] Through the Q-Learning algorithm, the PMIC can effectively learn how to adjust the power supply strategy under different conditions to minimize energy consumption while ensuring the normal operation of various devices in the tunnel. This process continuously optimizes the power management strategy, making the system more intelligent and adaptive in practical applications.

[0081] According to step S200, in the non-task or standby state, the system's PMIC puts the system into a low-power mode, turns off or reduces the power supply of unnecessary modules to minimize energy consumption; when the preset conditions are met, an interrupt signal is triggered, and the preset conditions include the sensor detecting a specific signal, an external input request, and the expiration of a timer; the interrupt signal interrupts the current low-power mode of the system and wakes up the processor; after the system responds to the interrupt signal, the processor enters the interrupt service program, which is responsible for handling the specific event that triggers the interrupt, and the system can complete data collection and data processing in the ISR; when the interrupt service program is executed, the system will return to the previous low-power mode; the PMIC will turn off the power supply of modules that are no longer needed, and the processor will enter a sleep or standby state, waiting for the next interrupt trigger.

[0082] According to step S200, the data collected by the sensor contains redundant information and noise. The lightweight algorithm preprocesses the raw data, including denoising, normalization and feature extraction, and uses a moving average algorithm for denoising. In the data screening stage, the threshold filtering algorithm and the differential algorithm are used to classify the data according to preset conditions or thresholds. Data that meet specific conditions will be retained, and the remaining data will be marked as non-essential data and discarded. The specific conditions represent abnormal data that exceeds the set range or key data within a specific time period. For the data of the temperature and humidity sensors, fluctuations within the normal range are regarded as non-essential data and are discarded. When the temperature or humidity exceeds the set range, the data will be retained and transmitted. For the monitoring data of the operating status of certain equipment in the cable tunnel, when the equipment is working in a normal state, the generated data is discarded; when the equipment is abnormal, the generated data will be retained.

[0083] The threshold filtering algorithm retains data that exceeds the threshold range and discards data within the normal range by setting one or more thresholds; the differential algorithm screens out data points with significant changes by calculating the difference between the current data and the data at the previous moment, and retains the data with significant changes. The threshold for determining whether the change is significant is determined by pre-configured content.

[0084] According to step S300, the sensor calibration value refers to the parameter obtained by calibrating the sensor according to the standard reference environment during the production process, which is used to ensure the measurement accuracy of the sensor in actual use; in the production workshop, the environmental conditions of the cable tunnel are simulated to calibrate the sensor; according to the calibration result, the calibration parameters of the sensor are adjusted using a software tool, and for different sensor types, the relationship between the sensor output and the actual physical quantity is modeled by a linear regression algorithm to determine the calibration value; the temperature sensor is calibrated by a standard source with a known temperature to obtain the calibration value;

[0085] The data processing threshold is a parameter used by the system to judge and filter data during operation, which determines when the system takes further processing or alarm operations on the collected data. In the production stage, the threshold range of different types of data is determined based on the historical data of the cable tunnel or the results of environmental simulation. The threshold is written into the system configuration file or firmware using software tools. The system can judge the validity or abnormality of the data during actual operation based on the pre-configured threshold.

[0086] According to step S400, the input features of the decision tree include the current status and environmental parameters of each device in the cable tunnel. The input features are used to construct the decision tree model, including: the battery power of the device , the current task load of the device , Equipment ambient temperature , the current working mode of the device and device task priorities ;

[0087] Each decision node makes a binary judgment based on a feature to decide which child node to move to. The input feature to Corresponding to nodes 1 to 5 respectively, each node determines whether the corresponding input feature is higher or lower than the set threshold, and the judgment is performed from node 1 to node 5 in sequence;

[0088] The leaf node represents the final power supply strategy decision. The action options of the leaf node include shutting down the device, reducing the power supply, and maintaining the power supply. If the battery power of the device is lower than the set threshold, and the current task load of the device is lower than the set threshold, and the ambient temperature of the device is lower than the set threshold, the device is shut down to save power; if the battery power of the device is lower than the set threshold but the task load still exists, the power supply voltage is reduced; if the current task load of the device is higher than the set threshold or the device is executing a high-priority task, the current power supply state is maintained.

[0089] According to step S400, historical data is used to mark the power supply strategy of each device in different states, so as to train the decision tree model; the best splitting feature of each decision node is selected by using information gain, and the feature space is recursively split until the preset node purity is reached; according to the current state of the device, the optimal power supply strategy is obtained through the decision tree model; in order to prevent the decision tree model from overfitting, the tree is pruned and unnecessary branches are removed; according to the feedback data in actual operation, the decision tree model is continuously updated and optimized.

[0090] A low-power integrated data processing system based on a cable tunnel, comprising:

[0091] Battery monitoring and low-power trigger module: including: battery power monitoring unit, low-power trigger unit, PMIC control unit and reinforcement learning power optimization unit; the battery power monitoring unit uses a battery power monitoring device to monitor the battery status of each device in the cable tunnel in real time, the low-power trigger unit determines the remaining battery power in combination with the battery life, and triggers the low-power mode when the power drops to the set threshold. The PMIC control unit controls the power supply of the cable tunnel through the power management integrated circuit PMIC, and the reinforcement learning power optimization unit uses the reinforcement learning algorithm to adjust the power supply of each device in the cable tunnel according to demand, and enters the low-power mode when there is no task;

[0092] Event processing and data transmission optimization module: including: interrupt drive unit, data processing and screening unit and LoRa communication unit; the interrupt drive unit processes events by interrupt drive, wakes up the system through the interrupt mechanism when data is collected or important events occur, and enters low power mode after processing. The data processing and screening unit uses a lightweight algorithm to perform preliminary data processing and screening, discards unnecessary data, and reduces the power consumption of data transmission and storage. The LoRa communication unit uses the low-power wireless communication protocol LoRa for data transmission, further reducing power consumption;

[0093] Parameter prefabrication and configuration module: including: parameter prefabrication unit and parameter configuration unit; wherein, the parameter prefabrication unit prefabricates the key parameters of the system through software tools during the manufacturing stage, and the key parameters are pre-configured by the parameter configuration unit according to the specific environment and application requirements of the cable tunnel, and the pre-configured contents include sensor calibration values ​​and data processing thresholds;

[0094] Low power strategy decision module: including: a decision tree analysis unit and a low power strategy output unit; wherein, the decision tree analysis unit uses a decision tree algorithm to analyze the status information of each device in the tunnel to determine whether to shut down or reduce the power supply of certain devices to reduce overall energy consumption, and the low power strategy output unit outputs the power supply strategy of the low power mode, which is applied to the low power mode of the PMIC.

[0095] In this embodiment, there are 5 devices in the cable tunnel, namely devices A, B, C, D, and E, which perform different tasks, such as security monitoring, communication, data acquisition, etc. The collected data is used as a training data set to establish a decision tree model. The training set contains the input features of each device in different states, including device status, environmental status, and corresponding output labels, which are represented as the optimal power supply strategy. The split nodes of the decision tree are determined according to the characteristics, such as device status, battery power, etc., and the leaf nodes output specific low-power strategies. The decision tree analysis unit receives real-time device status information and analyzes this information through the trained decision tree model to decide whether to shut down or reduce the power supply of the device. The low-power strategy output unit generates a specific low-power power supply strategy based on the output results of the decision tree and passes it to the PMIC for execution.

[0096] Devices A and D are in the Active state and their tasks are more critical, so the decision tree assigns them a higher low-power level. The decision tree recommends Low Power 1, which is about 4W and 5W. Devices B and E are in the Idle state, so the decision tree assigns them a lower power level. The decision tree recommends Low Power 2 and Low Power 3, which are about 3W and 2W, respectively. Device C is in the Sleep state and has the lowest power demand, so the power level assigned is the lowest. The decision tree recommends Low Power 3, which is about 2W.

[0097] The decision tree algorithm can reasonably allocate specific power consumption in low-power mode according to the different states and environments of the devices, achieving a dynamic balance of energy consumption. When all devices are running in low-power mode and have different specific power consumption levels, the overall energy consumption of the system is reduced by about 15% compared to the traditional fixed low-power mode.

[0098] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0099] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A low-power integrated data processing method based on cable tunnel, characterized in that: The steps include: S100, using a battery power monitoring device to monitor the battery status of each device in the cable tunnel in real time, and judging the remaining battery power in combination with the battery life, and triggering a low power consumption mode when the power drops to a set threshold; The power supply of the cable tunnel is controlled through the power management integrated circuit PMIC, and the power supply of each device in the cable tunnel is adjusted according to the demand using the reinforcement learning algorithm, and the low power consumption mode is entered when there is no task. According to step S100, the PMIC is responsible for unified management of the power supply of each device. The PMIC uses a reinforcement learning algorithm to establish a reinforcement learning model and dynamically adjust the power supply strategy according to the operating status and actual needs of different devices in the tunnel; The reinforcement learning model includes a state space, an action space, and a reward function, wherein the state space represents the system state that the PMIC can observe, including the operating state of each device, the battery state, and the environmental state; S200, process events in an interrupt-driven manner. When data is collected or an important event occurs, the system is awakened through the interrupt mechanism, and enters a low-power mode after the processing is completed; a lightweight algorithm is used for preliminary data processing and screening, and unnecessary data is discarded to reduce power consumption for data transmission and storage; a low-power wireless communication protocol LoRa is used for data transmission to further reduce power consumption; S300, in the manufacturing stage, the key parameters of the system are prefabricated through software tools. The key parameters are pre-configured according to the specific environment and application requirements of the cable tunnel. The pre-configured contents include sensor calibration values ​​and data processing thresholds; S400, using a decision tree algorithm, by analyzing the status information of each device in the tunnel, determining whether to shut down or reduce the power supply of certain devices to reduce overall energy consumption, outputting a power supply strategy for a low power mode, and applying it to the low power mode of the PMIC.

2. A low-power integrated data processing method based on a cable tunnel according to claim 1, characterized in that: According to step S100, the battery power monitoring device is responsible for continuously monitoring the battery status of each device in the cable tunnel, collecting the output voltage of the battery through the analog-to-digital converter, and the voltage gradually decreases as the battery discharge process proceeds; the current flowing through the battery is collected through the current sensor, and the actual discharge amount of the battery in a given time period is calculated by the coulomb counting method; The monitoring device records the initial capacity of the battery. As the battery is used, the monitoring device adjusts the actual available capacity of the battery by monitoring the capacity decay during the charge and discharge process. The monitoring device calculates the remaining capacity using the coulomb counting method based on the current voltage, current and the actual available capacity of the battery. When the battery power drops to a preset warning threshold, the battery power monitoring device generates a status feedback signal and sends the status feedback signal to the power management integrated circuit PMIC to trigger a low power consumption mode.

3. A low-power integrated data processing method based on a cable tunnel according to claim 2, characterized in that: The PMIC monitors the operating status of the equipment in the tunnel and identifies the current working mode. The PMIC evaluates the power demand of each device based on the power supply strategy and determines the required power supply level; Based on the evaluation results of the device requirements, the PMIC adjusts the output voltage and current; the PMIC provides corresponding power support when the device has actual tasks, and during non-task time periods, the PMIC turns off the power supply of the module that provides the corresponding device and enters low-power mode; PMIC assigns power supply priorities to different devices. There are several power inputs in the cable tunnel, and PMIC is responsible for switching between power supplies. PMIC has redundant power management capabilities. When a power supply fails, PMIC can switch to another power path to ensure uninterrupted power supply to the device.

4. The low-power integrated data processing method based on cable tunnel according to claim 3 is characterized in that: The operating state includes working, standby and dormant; the battery state includes power percentage, voltage and current; the environmental state includes temperature and humidity; the action space represents the actions that the PMIC can take, including high power supply, low power supply and power off, and the high power supply and low power supply are specifically adjusted according to the state space; the reward function is used to measure the effect of the action taken by the PMIC, and the reward value is associated with the energy consumption and performance goals of the system; The Q-Learning algorithm is used as a reinforcement learning algorithm to determine the strategy by iteratively updating a table called the Q value. The formula is as follows: ; in, is in state Take Action The Q value at the time represents the execution action The expected cumulative reward after is the learning rate, ranging from , control the degree to which new information updates old information; Is the immediate reward at the current moment, reflecting the action effect; is the discount factor, ranging from , weighing the relative importance of future rewards versus current rewards; Indicates that in the next state In the example above, choose the action that brings the greatest reward. The corresponding Q value; PMIC continuously monitors the real-time status of the equipment in the tunnel and uses the real-time status as the input S of the reinforcement learning algorithm. At each time step t, PMIC And the learned Q value, select an optimal action ; When the PMIC performs an action, the system gives an instant reward based on the current energy consumption and device performance. ,When the PMIC successfully reduces unnecessary power supply and the device still operates normally, the system gives a positive reward; if the insufficient power supply causes the device to degrade, a negative reward is given; PMIC continuously updates the Q value table through the Q-learning formula and gradually learns the optimal power management strategy under various conditions.

5. The low-power integrated data processing method based on cable tunnel according to claim 1, characterized in that: According to step S200, in a non-task or standby state, the PMIC of the system puts the system into a low power consumption mode, shuts down or reduces the power supply of unnecessary modules to minimize energy consumption; when a preset condition is met, an interrupt signal is triggered, and the preset condition includes a sensor detecting a specific signal, an external input request, and a timer expiration; the interrupt signal interrupts the current low power consumption mode of the system and wakes up the processor; After the system responds to the interrupt signal, the processor enters the interrupt service routine, which is responsible for handling the specific event that triggers the interrupt. The system can complete data collection and data processing in the ISR; when the interrupt service routine is executed, the system will return to the previous low-power mode; the PMIC will turn off the power supply of modules that are no longer needed, and the processor will enter sleep or standby state, waiting for the next interrupt trigger.

6. A low-power integrated data processing method based on a cable tunnel according to claim 5, characterized in that: According to step S200, the data collected by the sensor contains redundant information and noise, and the lightweight algorithm performs denoising by preprocessing the raw data, including denoising, normalization and feature extraction, and uses a moving average algorithm for denoising; In the data screening stage, the threshold filtering algorithm and the differential algorithm are used to classify the data according to the preset conditions or thresholds. The data that meet the specific conditions will be retained, and the rest will be marked as non-essential data and discarded. The specific conditions represent abnormal data that exceeds the set range or key data within a specific time period; for the data of the temperature and humidity sensors, the fluctuations within the normal range are regarded as non-essential data and discarded, and the data will be retained and transmitted when the temperature or humidity exceeds the set range; for the monitoring data of the operating status of certain equipment in the cable tunnel, when the equipment is working in a normal state, the generated data is discarded; when the equipment is abnormal, the generated data will be retained; The threshold filtering algorithm retains data that exceeds the threshold range and discards data within the normal range by setting one or more thresholds; the differential algorithm screens out data points with significant changes by calculating the difference between the current data and the data at the previous moment, and retains the data with significant changes. The threshold for determining whether the change is significant is determined by pre-configured content.

7. The low-power integrated data processing method based on cable tunnel according to claim 1, characterized in that: According to step S300, the sensor calibration value refers to the parameter obtained by calibrating the sensor according to the standard reference environment during the production process, which is used to ensure the measurement accuracy of the sensor in actual use; in the production workshop, the environmental conditions of the cable tunnel are simulated to calibrate the sensor; According to the calibration results, use software tools to adjust the calibration parameters of the sensor. For different sensor types, use linear regression algorithms to model the relationship between the sensor output and the actual physical quantity to determine the calibration value. Calibrate the temperature sensor using a standard source with a known temperature to obtain the calibration value. The data processing threshold is a parameter used by the system to judge and filter data during operation, which determines when the system takes further processing or alarm actions on the collected data; During the production phase, the threshold ranges for different types of data are determined based on the historical data of the cable tunnel or the results of environmental simulation. The thresholds are written into the system's configuration files or firmware using software tools. The system can determine the validity or anomalies of the data during actual operation based on the pre-configured thresholds.

8. The low-power integrated data processing method based on cable tunnel according to claim 1, characterized in that: According to step S400, the input features of the decision tree include the current status and environmental parameters of each device in the cable tunnel. The input features are used to construct the decision tree model, including: the battery power of the device , the current task load of the device , Equipment ambient temperature , the current working mode of the device and device task priorities ; Each decision node makes a binary judgment based on a feature to decide which child node to move to. The input feature to Corresponding to nodes 1 to 5 respectively, each node determines whether the corresponding input feature is higher or lower than the set threshold, and the judgment is performed from node 1 to node 5 in sequence; The leaf node represents the final power supply strategy decision. The action options of the leaf node include shutting down the device, reducing the power supply, and maintaining the power supply. If the battery power of the device is lower than the set threshold, and the current task load of the device is lower than the set threshold, and the ambient temperature of the device is lower than the set threshold, the device is shut down to save power; if the battery power of the device is lower than the set threshold but the task load still exists, the power supply voltage is reduced; if the current task load of the device is higher than the set threshold or the device is executing a high-priority task, the current power supply state is maintained.

9. A low-power integrated data processing method based on a cable tunnel according to claim 8, characterized in that: According to step S400, historical data is used to mark the power supply strategy of each device in different states, so as to train the decision tree model; the best splitting feature of each decision node is selected by using information gain, and the feature space is recursively split until the preset node purity is reached; according to the current state of the device, the optimal power supply strategy is obtained through the decision tree model; in order to prevent the decision tree model from overfitting, the tree is pruned and unnecessary branches are removed; according to the feedback data in actual operation, the decision tree model is continuously updated and optimized.

10. A low-power integrated data processing system based on a cable tunnel, using a low-power integrated data processing method based on a cable tunnel according to any one of claims 1 to 9, characterized in that: include: Battery monitoring and low-power trigger module: including: battery power monitoring unit, low-power trigger unit, PMIC control unit and reinforcement learning power optimization unit; the battery power monitoring unit uses a battery power monitoring device to monitor the battery status of each device in the cable tunnel in real time, the low-power trigger unit determines the remaining battery power in combination with the battery life, and triggers the low-power mode when the power drops to the set threshold. The PMIC control unit controls the power supply of the cable tunnel through the power management integrated circuit PMIC, and the reinforcement learning power optimization unit uses the reinforcement learning algorithm to adjust the power supply of each device in the cable tunnel according to demand, and enters the low-power mode when there is no task; Event processing and data transmission optimization module: including: interrupt drive unit, data processing and screening unit and LoRa communication unit; the interrupt drive unit processes events by interrupt drive, wakes up the system through the interrupt mechanism when data is collected or important events occur, and enters low power mode after processing. The data processing and screening unit uses a lightweight algorithm to perform preliminary data processing and screening, discards unnecessary data, and reduces the power consumption of data transmission and storage. The LoRa communication unit uses the low-power wireless communication protocol LoRa for data transmission, further reducing power consumption; Parameter prefabrication and configuration module: including: parameter prefabrication unit and parameter configuration unit; wherein, the parameter prefabrication unit prefabricates the key parameters of the system through software tools during the manufacturing stage, and the key parameters are pre-configured by the parameter configuration unit according to the specific environment and application requirements of the cable tunnel, and the pre-configured contents include sensor calibration values ​​and data processing thresholds; Low power strategy decision module: including: a decision tree analysis unit and a low power strategy output unit; wherein, the decision tree analysis unit uses a decision tree algorithm to analyze the status information of each device in the tunnel to determine whether to shut down or reduce the power supply of certain devices to reduce overall energy consumption, and the low power strategy output unit outputs the power supply strategy of the low power mode, which is applied to the low power mode of the PMIC.

Images